International Archives of the Photogrammetry, Remote Sensing and Spatial Information Science, Volume XXXVIII, Part 8, Kyoto Japan 2010
REMOTE SENSING FOR DISSASTER MITIGATION: CASE STUDY FOR TSUNAMI EVACUATION ROUTE
MODELLING IN CILACAP-CENTRAL JAVA, INDONESIA
Ratna Sari Dewia*, Niendyawati Salama, Suwahyuono Suwadib.
a
Researcher at Center for Marine Resources Surveys-National Coordinating Agency for Surveys and Mapping.
b
Head of Center for Marine Resources Surveys-National Coordinating Agency for Surveys and Mapping.
KEY WORDS: Remote sensing, Mitigation, Tsunami Evacuation Model, Evacuation Route, Cilacap
ABSTRACT
Indonesian region is located in the interaction between three major plates, one of the most active subduction zones. This region has
frequent earthquakes and tsunamis. Given its location, Indonesian region is very vulnerable to many disasters, including tsunami.
The rapid growth of population in coastal areas makes Indonesia more vulnerable to tsunami occurrence since coastal areas have
always been the most preferred location for settlements. This research was conducted in Cilacap Regency, Indonesia which was hit
by tsunami on July 17th, 2006, caused many damages and casualties.
In the tsunami mitigation plan, evacuation plays a crucial measure for saving human lives. The primary strategy for saving lives
immediately before tsunami waves arrives is to evacuate people from the hazard zone. Especially for communities who are living in
a low-lying coastal area. Given their location, it would be impossible to evacuate these communities in time, which could result in
significant loss of life.
This research tried to determine the suitable location for vertical evacuation and the capacity of suitable evacuation shelter buildings
and also to develop a methodology to choose the most effective evacuation routes. For these purposes, some kind of spatial
information are required such as the distribution pattern of settlements and detailed road network, in which for these study, can
significantly be supported by using remote sensing technology. This paper will present the results of how remote sensing
technologies were used to support provision of spatial information required for tsunami evacuation model in Cilacap, Central Java.
* Corresponding Author.
1.
INTRODUCTION
In the case of buildings for evacuation, because tsunami has
long period to be occurred, there are no certain buildings
functioning only as evacuation shelter buildings. Finding the
suitable buildings that can be used as evacuation shelter
building is necessary due to high building construction cost and
also efficiency in space occupation. A mosque, school, high
floored-house and etc. can be used as evacuation shelter
buildings. These evacuation shelter buildings should have
certain design and configuration including size, shape and
orientation (NTHMP, 2001).
Coastal areas have always been the most preferred location for
settlements. It is because the attractiveness of coastal areas and
many economic opportunities providing by those areas.
Because of these factors and also the long gaps between
tsunami destructing events, coastal people tend to develop new
housing, marine facilities, and resort developments. These
make people and facilities are threatened by the destructiveness
of tsunami. Additionally, given their location, it would be
impossible to evacuate these communities in time, which could
result in significant loss of life. In this condition, evacuation
planning are considered as instrument for risk mitigation at the
local scale, and it is commonly by choosing existing roads as
suggested evacuation routes.
It is important to identify to which evacuation building should
people in a certain area go and how to get there in a fast and
efficient way. It is important to identify the existing buildings
that are potential to function as evacuation shelter building, and
it should be integrated with the effort of finding new evacuation
shelter building locations. Furthermore, the existence of
evacuation shelter building should be included in spatial plan
with regard to disaster mitigation aspects.
Based on the sources and time to travel, tsunami can be
classified as local and distant. Local tsunami can arrive at
nearby shores only within minutes, for example in 17 July 2006
tsunami occurrence in Cilacap, the first wave of tsunami struck
its coastal area one hour after the earthquake and can have a
huge damage resulting from triggering earthquake such as
ground shaking, landslides and surface faulting, etc. On the
other hand, distant tsunami can travel for hours before hitting a
coastline. These types of tsunami will determine the time
needed to respond and act on warnings. In Indonesia most of
tsunami events are local so that the tsunami warning should be
announced in a very short time. For this case, early warning
system would be the critical point for evacuation route
determination and evacuation planning as well.
The objective in this study was to examine the potential of
satellite remote sensing in supporting tsunami mitigation
planning in the Central Java, Indonesia. Remote sensing data
and methods were used to support the development of tsunami
evacuation maps for the coastal areas of Cilacap such as for
identifying buildings as the sources of evacuee, existing
building for evacuation and also to improve the existing road
network. Geographical Information Systems (GIS) was used to
manage all spatial data and to create and use all components in
the tsunami evacuation model.
281
International Archives of the Photogrammetry, Remote Sensing and Spatial Information Science, Volume XXXVIII, Part 8, Kyoto Japan 2010
2.
There are many factors to be considered in selecting suitable
method in tsunami evacuation, as follows:
TSUNAMI EVACUATION MODEL
The principle function of evacuation is to make sure that people
move from a relatively danger place to a safer place via a route
that is itself free from significant danger (Webb, 2005).
Furthermore, the main strategy of evacuation is to save lives.
NTHMP (2001) explained that there are two methods of how to
evacuate people from tsunami hazard zones to the safer place,
as follows; a). Horizontal evacuation; people move from the
hazard zones to the safer areas in a distant locations or higher
ground such as hills; b). Vertical evacuation; this method will
evacuate people to the higher floors of a tsunami-resistant
building nearby.
Applications that can create an efficient evacuation plan are
needed to help evacuate community to safer areas in case of
disaster. Huang et. al. (2005) stated that evacuation planning
tools can be categorized into simulation-based approaches to
which describes the responses and route behaviour of evacuees
including estimating the clearance time, etc and the second one
is analytical approaches which can optimize the evacuation
process including roads, shelters, rescue points, etc. Further,
many researcher try to combine both approaches.
Figure 1. Decision-making and design process for vertical
evacuation structure (Source : FEMA, (2008b))
Silva et.al. (2003) explained that to study an evacuation system,
two technologies are used namely, simulation modeling and
Geographical Information Systems (GIS). A GIS is able to store
spatially referenced data and information and provide
sophisticated mapping and display facilities. Many existing GIS
come with built-in analysis tools in addition to their data
storage, manipulation and representation tools. The problem
and decision making scenario provide an interacting elements
that have a spatial dimension which can be associated to
geographical coordinates of the Earth’s surface.
For
illustration, location elements such as shelters, population
generation points (houses, workplaces, schools, hospitals, etc),
hazard locations, etc., and directional elements such as
travelling vehicles, wind and other meteorological features, etc.
all can be linked to a spatial dimension.
a) Evacuation time
The available time for evacuation mostly depend on the
remaining time between tsunami warning alarms to the arrival
of tsunami wave. The evacuation time can vary according to the
type of tsunami. It varies from several minutes to hours. After
an earthquake generated tsunami, early warning system will
take several minutes to analyze and decide whether this
earthquake will cause tsunami and disseminate the information
to the authority. Horizontal evacuation is better applied in the
area which is prone to distant tsunami due to longer tsunami
arrival time, so that there will be more time to move to the safer
place in a distant or higher ground. On the other hand, vertical
evacuation will be suitable for the area exposed to local
tsunami. It is because the tsunami waves will arrive on shore in
a shorter time.
b) Topography
Topography characteristics such as elevation will determine
also the method of evacuation to be implemented. High
elevation of land which is higher than tsunami waves like hill
or high inland is required by horizontal evacuation. Meanwhile,
coastal area which has low-lying area and maybe there is no hill
nearby will require vertical evacuation.
c) Wave run-up and tsunami inundation
Coastal characteristics, source of tsunami, magnitude and the
existing of coastal feature such as mangrove and coastal forest
will determine the height of run-up and the extent of inundation
area. Vertical evacuation is required when run-up and
inundation level grow faster due to for example the bigger
magnitude of earthquake, whilst, horizontal evacuation is
needed when run-up and inundation level grow slowly.
d) Supporting facilities and infrastructure
Roads, bridges and evacuation route are needed to support the
movement of huge number of population within the same time
period. In addition, a public room in the evacuation place and
also mode of transportation are necessary to be considered.
Moreover, vertical evacuation needs buildings which are
adequate in terms of capacity, strength, utility and also
capability. For those methods of evacuation (Table 1), an
Despite providing a helpful mechanism to deal with problem
which have spatial dimension, a GIS cannot provide the
complex decision model required to overcome real-world
problem (Church et. al. (2000) after Silva et. al. (2003)). In this
case, simulation modeling is essential to develop an appropriate
tool to handle the spatial decision making aspects of the
evacuation planning process. Church and Sexton (2002)
concluded that most of the research has been concentrated on
two distinct problems, evacuation of buildings and evacuation
of large areas, like entire cities or coastal plains.
Figure 1 from FEMA (2008b) explains the process of decision
making and design of vertical evacuation in particular area. In
fact, vertical evacuation is needed when there is not enough
time between warning and tsunami inundation to allow a
community to evacuate inundation zone or to higher ground. In
most cases this will be communities at risk for near-sourcegenerated tsunamis.
The implementation of those methods in the prone area to
tsunami depends upon the characteristic of tsunami and
location, topography, existing facilities and evacuation time.
282
International Archives of the Photogrammetry, Remote Sensing and Spatial Information Science, Volume XXXVIII, Part 8, Kyoto Japan 2010
effective warning systems and public information, notification,
and training program are critical to the success of all evacuation
measures.
building maps can be derived from high resolution images such
as IKONOS, QuickBird, etc. High resolution images are needed
when the more detail data regarding population for each
building are required. In fact, building inventories for tsunami
evacuation model can be carried out at various scale levels,
depending on the requirements of the tsunami evacuation
model. There are many different scales have been identified,
ranging from small scale to detailed scale (Table 3).
Table 1. Suitability of tsunami evacuation method
Evacuation Method
Tsunami travel time
Location &
Topography
Wave run-up &
inundation level
Required facilities &
infrastructure
Horizontal
Long (hours)
Close to hills or
higher elevation
ground
Slow run-up,
inundation <1 m
Evacuation route,
road & bridges,
mode of transport
Vertical
Short (minutes)
Flat, low-lying
land
Table 2. Examples of remote sensing application in each risk
management phase
Fast run-up,
inundation >1 m
Multi-storey
building, escape
hill
Prevention
Earthquake
Source : Budiarjo (2006)
Volcanic eruptions
Landslides
In this research, the focus will be on the building for vertical
evacuation of tsunami. Evacuation is not a new concept for
emergency planners. Even though, the purpose of evacuation is
to move people away from an actual or potential danger to a
safer place, it is not an easy option and may not be the safest
option as well (HMG, 2006). So evacuation should not be
automatically to be adopted. Simulation can be define as the
process of designing a model of a real system and conducting
experiments with this model for the purpose of understanding
the behavior of the system and /or evaluating various strategies
for the operation of the system (Shannon, 1998). These
assumption falls into two categories: structural assumptions
which involve simplification and abstraction of reality on issues
involving the operation and the processes of the system, and the
data assumption which are based on the collection of reliable
data and its proper analysis.
3.
Flash
floods
Major
floods
Storm
surge
Mapping
global
lineaments
and landuse
Topographic
and landuse
maps
Topographic
and landuse
maps
Landuse
maps
Flood plain
maps,
landuse maps
Landuse and
land cover
maps
Hurricanes
Tornadoes
Drought
Tsunamis
DESIGN OF REMOTE SENSING DATA TO
SUPPORT EVACUATION MODEL
Inundation
map, hazard
maps and
landuse map
Preparedness
(warning)
Geodynamic
measurements of
strain accumulation
Relief
Locate stricken
areas, map damage
Detection/measureme
nt of gaseous
emission
Soil porosity: rainfall,
slope stability
Mapping lava
flows, ashfalls and
lahars, map damage
Mapping slide area
Local rainfall
measurements
Regional rainfall,
evapo-transpiration
Map flood damage
Sea state, ocean
surface wind
velocities
Synoptic weather
forecast
Now casts, local
weather observations
Long-ranged climate
model
Map extent of
damage
Evacuation route
maps
Map extent of
floods
Map extent of
damage
Map amount, extent
of damage
Monitoring
vegetative biomass;
station communication
Map extent of
damage
Source: Walter (2008)
Note:
Normal
: Operational or needs very little research
Underlined : Research and development required
Bold
: Requires improved observation capability
Italics
: Requires improved spatial or temporal resolution
Remote sensing is one of the most important tools in risk
management, from modeling and vulnerability analysis, from
early warning, to damage assessment. It is very useful in the
planning process in general, and also valuable in detecting and
mapping many types of natural hazards when detailed
descriptions of their effects do not exist. The development of
remote sensing technologies has improved the mapping abilities
and increased its applications. Some examples of remote
sensing application in three phases of risk management are
presented in Table 2.
Table 3. Building inventory versus mapping scale
Scale of analysis
Small : < 1:100.000
Medium : 25-50.000
It is very essential to know what types of remote sensing
information are suitable for identifying and assessing particular
natural hazards. Furthermore, an effective utilization of remote
sensing data depends on the capability of the users in accurately
interpreting photographs, images, graphs or statistics which are
derived from remotely sensed data. There are many factors
which determine the uses of remote sensing data in natural
hazard assessment, such as scale, resolution, and tone or color
contrast. Moreover, coverage, frequency, data cost and
availability are also incorporated.
Large : 10.000
Detailed : >1:10.000
In case of tsunami evacuation model, more detail data regarding
buildings are needed, since the evacuation model is started from
the concentration of evacuees. From building maps we can have
building footprints represented the residential of evacuees. The
Building inventory
By Municipality
• Nr. Buildings
Mapping units
• Predominant type ( e.g. residential,
commercial, industrial)
• Nr. Buildings
Building footprints
• Generalized use
• Height
• Building types
Building footprints
• Detailed use
• Height
• Building types
• Construction type
• Quality / Age
• Foundation
Modified from (Alkema et. al., 2009)
283
International Archives of the Photogrammetry, Remote Sensing and Spatial Information Science, Volume XXXVIII, Part 8, Kyoto Japan 2010
4.
d.
DATA AND METHOD
e.
4.1 Study Area
Mosques are identified from building orientation to Qiblat
direction that differs from surrounding buildings and also
by rectangular shape of the roof.
Warehouse is identified through large roof plane and large
open space surrounded by commercial facilities
Cilacap is one of the regencies in Central Java Province,
Indonesia. Some part of its region is protected by Nusa (island)
Kambangan and the others are directly facing the Indian Ocean
(Figure 2).
Figure 2. Overview of study areas
Some areas in Cilacap have become industrial centers and sea
ports which by its nature making these areas highly populated.
Many buildings are situated on the coastal plain just a few
meters away from the present shoreline since the region
occupies a flat area, some of which are fishing village forming
squatter areas. Although some parts of Cilacap are protected by
Nusa Kambangan but there have been many damages and
casualties due to south Java tsunami occurrence on July 17th,
2006. The disaster took hundreds of lives, devastated the beach,
damaged hundreds of fishing boats and houses, etc
(BAKORNAS PBP, 2006; Mori et. al., 2007).
Figure 3. Building identification from QuickBird image
(modified from Budiarjo, 2006)
Ruko (“rumah-toko”, shop in ground floor and residential use at
the upper floor) is identified from the rectangular multi-storey
building mass with short flank heading the main road. The roof
of ruko is mostly concrete flat deck with small part covered
with tile or corrugated metal sheet.
Further, field observation was conducted to validate the
building use maps and to assess buildings for ESB. The
sampling design was determined based on tsunami run-up and
inundated area. Hence the building site was classified into: seafront site, river/canal-front site, and tsunami reached inland site.
Buildings to be checked were selected purposively based on
their uses in each building site and also based on preliminary
survey having conducted. Despite based on their building sites,
the selection of building samples were also determined by the
diversity of building uses. The sites which have a high diversity
of building uses had more building samples taken. For example,
the area which has many facilities such as office, school, or
other type of building uses, will have more samples to be
observed. Buildings were classified based on their uses by
knowing the characteristic of the objects in the image. For these
purposes, buildings were classified into several classes, such as
house, office, mosque, church, school, shop, hotel, sport-center,
factory, and fishing market.
4.2 Method
Building Identification; Tsunami inundation of supplied
model from worst scenario was used as a basis in selecting the
locations which require vertical tsunami evacuation. The
building map was extracted from 0.6 m resolution of QuickBird image. The buildings serve as the source of evacuees. The
building maps were then classified based on their building uses.
Heads up digitizing method was used to delineate each building
block by using visual interpretation. In this case, professional
knowledge and experience combined with the interpretation
element were needed in interpreting the image. The large
buildings could be orange to brown (depending on roof style)
and their locations in relatively easy-accessed areas. In detailed,
Budiarjo (2006) and Janssen (2004) stated that the
identification of the facilities can be illustrated as follows
(Figure 3).
a.
b.
c.
Several requirements of building resistance for tsunami to
conduct building assessment were as follows: 1) Located at a
distance of more than 200 m from shoreline or 100 m from a
river near the coast; 2) Located near the population
concentration; 3) Having alternate function as mosque, school,
parliament building, government office, market, shopping
centre, convention centre, sport hall, hotel, and parking
building; 4) Building floor reserved for evacuation located
above tsunami wave height in the area; 5) Well-planned and
designed; 6) Good quality construction (tsunami and
earthquake resistant building).
A U shape building or L shape building with open field
can be school because of its location near settlement.
The offices are identified through their building mass
typology.
The existence of certain facilities such as market building,
port, bus station can be easily identified from field
observation and from image.
284
International Archives of the Photogrammetry, Remote Sensing and Spatial Information Science, Volume XXXVIII, Part 8, Kyoto Japan 2010
Road Network Identification: The other component for
tsunami evacuation model is road network. It is one of the main
components of the model since network analyst is used for
generating the evacuation route. The network model is made up
of arc segments with appropriate network connectivity settings
and attributes defined. Networks typically have rules about how
objects move through them.
In order to perform analysis, a network datasets must be
developed on the available road network, otherwise it can be
derived from high resolution image also. The available road
network sometimes are not detailed enough to support the
model. In this case, high resolution images and field survey are
required to have a detailed road network include setting the
travel impedance for each segment, defining directions and
one-way streets, and also managing restricted turns.
Figure 4. The results of building identification using high
resolution image and field observation
The existing road networks were from topographic map of
BAKOSURTANAL scale 1:25.000. This data was in shape file
format. The road networks from Topographic map are less
detailed than that is needed for this research. Consequently, the
more detailed road networks were added from high resolution
image.
The overall processes were as follows:
1). Geometric correction of Quick-Bird image using
topographic map scale 1:25.000.
2). Overlaying the rectified image with existing road network
data (shape file of topographic map).
3). Adding the missing road using heads up digitizing method
in ArcGIS to make the available road networks more
detailed.
4). Giving attributes to the road data and classifying the road
into five classes based on the road width, i.e. arterial road,
collector road, local road, other road and pathway.
The detailed road network is necessary to be implemented in
the network modeling. Road network is very essential for
evacuation purposes since it will determine the movement of
evacuees. It also serves as connection to the shelters and
provides the evacuation routes itself. The accessibility to an
ESB will require sufficient evacuation route so that evacuees
can reach the ESBs in time.
These road networks were also added during fieldwork by
using mobile PDA and GPS. During fieldwork, many missing
roads were identified. Usually they were categorized as path
and were located in the middle of settlements and covered by
trees, making difficult to identify them from the image.
5.
RESULTS AND DISCUSSIONS
Figure 5. The improved road network
The building use classification is needed to be determined
because it gives contribution in evacuation process since it will
serve as target point of evacuations. The uses of the buildings
will determine the number of people who present in the
buildings in the different periods of time. For these purposes,
building classes were classified into house, office, mosque,
church, school, shop, hotel, sport-center, factory, and fish
market (Figure 4). From this classified building maps, there
were two objectives could be achieved. First, facilities location
could be identified. By knowing the location of facilities, the
estimation of population during daytime and night-time could
be made since the number of population in offices, schools and
other facilities could be estimated. In this case, a high
resolution image is needed to identify the type of the building
uses. Further this information is required to estimate the
number of population in each building. Second, from these
building maps, the potential existing shelter buildings for
tsunami evacuation could be recognized.
Figure 6. The available road network from topographic map
285
International Archives of the Photogrammetry, Remote Sensing and Spatial Information Science, Volume XXXVIII, Part 8, Kyoto Japan 2010
6.
FEMA (2008b). "Guidelines for Design of Structures for
Vertical Evacuation from Tsunamis." Retrieved 25 March
2009, 2009, from https://www.atcouncil.org/pdfs/
FEMAP646A.pdf.
CONCLUSION
The realistic model from tsunami evacuation model is
determined by the detailed input data of road network, network
attributes and population data. The detailed road network will
enhance the model since the travel will be carried on from the
centroids of tessellations (point of origin) to the closest
network. In line with developing more detailed road network,
more detailed network attributes will result in a better
evacuation model. Other thing which is important to improve is
source of evacuees. In this case the detailed building type is
necessary so that the number of evacuees could be estimated.
HMG, H. G. (2006). Evacuation and Shelter Guidance.
Easingwold - York, Crown.
Janssen, L. L. F. (2004). Visual Image Interpretation. Principles
of Remote Sensing. N. Kerle, L. L. F. Janssen and G. C.
Huurneman. Enschede, ITC: Page 177.
Mori, J., et al. (2007). "The July 17, 2006 West Java
Earthquake and Tsunami." Annuals of Disaster Pre. Res. Inst.
Kyoto Univ. 50 A.
ACKNOWLEDGEMENT
NTHMP (2001). Designing for Tsunami - Seven Principles for
Planning and Designing for Tsunami Hazards. National
Tsunami Hazard Mitigation Program. USA, NOAA, USGS,
FEMA, NSF, Alaska, California, Hawaii, Oregon, and
Washington.
Thanks to Bappenas and Netherlands Education Centre, for
providing a scholarship to reach higher education in Gadjah
Mada University (GMU) and ITC. I would also like to extend
my many thanks to Marine Natural Resources Survey Centre of
BAKOSURTANAL that allows me to study and improve my
knowledge. I am deeply indebted to my supervisors for their
supervision, encouragement and guidance throughout the
research; and also for their critical comments and helpful
guidance give me a chance to explore further.
Shannon, R. E. (1998). Introduction to the Art and Science of
Simulation. Winter Simulation Conference, USA.
Silva, F. N. d., et al. (2003). Chapter XXI: Evacuation Planning
and Spatial Decision Making: Designing Effective Spatial
Decision Support System through Integration of Technologies.
In Decision Making Support Systems. Decision Making
Support Systems: Achievements and Challenges for the New
Decade. M. Mora, G. Forgionne and J. N. D. Gupta. London,
IDEA Group Publishing: 358 pages.
REFERENCE
ADPC (2008). Total Disaster Risk Management Good Practices
2008. Kobe, Asian Disaster Reduction Center.
Webb, T. (2005). Review of New Zealand’s preparedness
tsunami hazard, comparison to risk and recommendations for
treatment, Institute of Geological and Nuclear Science.
Alkema, D., et al. (2009). Guide Book: Multi-hazard risk
assessment, United Nations University – ITC School on
Disaster Geoinformation Management (UNU-ITC DGIM).
BAKORNAS PBP (2006). Laporan Perkembangan Penanganan
Bencana Gempa Bumi dan Tsunami di Jawa Barat, Jawa
Tengah dan DI Jogjakarta.
Budiarjo, A. (2006). Evacuation shelter building planning for
tsunami prone area : a case study of Meulaboh city, Indonesia.
Enschede, ITC: 112 pages.
Church, R. L. and R. M. Sexton (2002). Modeling small area
evacuation: Can existing transportation infrastructure impede
public safety? Santa Barbara - California, Vehicle Intelligence
& Transportation Analysis Laboratory - University of
California. Final Report.
286